The R Project SVN – R-packages – Diff – /trunk/SQLiteDF/src/sqlite_dataframe.c


#include <stdio.h>
#include <string.h>
#include "sqlite_dataframe.h"
 
/****************************************************************************
 * UTILITY FUNCTIONS
 ****************************************************************************/
 
/* if user supplied a name (1st arg), return that name. otherwise, use the 
 * default "data". rname is R name, iname is internal name */
static int _check_sdf_name(SEXP name, char **rname, char **iname, int *file_idx) {
    int namelen = 0;
 
    /* check if arg name is supplied */
    if (name == R_NilValue) {
        *rname = "data";
        namelen = 5;
        *iname = (char*)R_alloc(13, sizeof(char)); /* .SQLiteDF/data10000.db\0 */
        *file_idx = 1;
        sprintf(*iname, "%s%d.db", *rname, *file_idx);
    } else if (IS_CHARACTER(name)) {
        *rname = CHAR_ELT(name,0);
        if (!_is_r_sym(*rname)) { 
            error("supplied name \"%s\"is not a valid R symbol.", *rname); 
 
 
        }
        namelen = strlen(*rname);
        *iname = (char*)R_alloc(namelen + 9, sizeof(char)); /* .SQLiteDF/<name>10000.db\0 */
        sprintf(*iname, "%s.db", *rname);
    } else {
        Rprintf("Error: the supplied value for arg name is not a string.\n");
    }
    return namelen;
}
 
static int _find_free_filename2(char *rname, char *dirname, char **iname, int *namelen, int *file_idx) {
    sqlite3_stmt *stmt;
    char *tmp_iname;
    sqlite3_prepare(g_workspace, "select 1 from workspace where internal_name=?", -1,
            &stmt, NULL);
    do {
        if (dirname != NULL) {
            sprintf(g_sql_buf[2], "%s/%s", dirname, *iname);
            tmp_iname = g_sql_buf[2];
        } else {
            tmp_iname = *iname;
        }
 
        if (!_file_exists(tmp_iname)) {
            sqlite3_reset(stmt);
            sqlite3_bind_text(stmt, 1, tmp_iname, -1, SQLITE_STATIC);
            if (sqlite3_step(stmt) != SQLITE_ROW) break;
        }
        *namelen = sprintf(*iname, "%s%d.db", rname, ++(*file_idx)) - 3;
    } while (*file_idx < 10000);
 
    sqlite3_finalize(stmt);
    return *file_idx;
}
 
/* creates an sdf attribute table */
static int _create_sdf_attribute2(const char *iname) {
    sprintf(g_sql_buf[2], "create table [%s].sdf_attributes(attr text, "
            "value text, primary key (attr))", iname);
    int res;
    res = _sqlite_exec(g_sql_buf[2]);
    if (res == SQLITE_OK) {
        sprintf(g_sql_buf[2], "insert into [%s].sdf_attributes values ('name',"
                "'%s');", iname, iname);
        res = _sqlite_exec(g_sql_buf[2]);
    }
    return res;
}
 
char *_create_sdf_skeleton1(SEXP name, int *onamelen, int protect) {
    char *iname, *rname;
    int namelen, file_idx = 0, res;
 
    namelen = _check_sdf_name(name, &rname, &iname, &file_idx);
 
    if (!namelen) return NULL;
 
    _find_free_filename2(rname, ".SQLiteDF", &iname, &namelen, &file_idx);
 
    if (file_idx >= 10000) { 
        Rprintf("Error: cannot find free SDF name.\n");
        return NULL;
    }
 
    /* add to workspace */
    sprintf(g_sql_buf[3], ".SQLiteDF/%s", iname);
    iname[namelen] = 0; /* remove ".db" */
    res = _add_sdf1(g_sql_buf[3], iname);
    if (_sqlite_error(res)) return NULL;
 
    /* detach SDF's if necessary to attach this one. if file does not
     * exist, then a sqlite db will be created after we make our 1st table */
    if (!USE_SDF1(iname, FALSE, protect)) { /* _delete_sdf2(iname); */ return NULL; }
 
 
    /* create attributes table */
    res = _create_sdf_attribute2(iname);
    if (_sqlite_error(res)) {
        sprintf(g_sql_buf[2], "detach '%s'", iname);
        _delete_sdf2(iname);
        return NULL; 
    }
 
    if (onamelen) *onamelen = namelen;
    return iname;
}
/* checks if a column has a corresponding factor|ordered table */
int _is_factor2(const char *iname, const char *factor_type, const char *colname) {
    sqlite3_stmt *stmt;
    sprintf(g_sql_buf[2], "select * from [%s].[%s %s]", iname, factor_type,
            colname);
    int res;
    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
    sqlite3_finalize(stmt);
    return res == SQLITE_OK;
}
 
/* create a factor|ordered table */
int _create_factor_table2(const char *iname, const char *factor_type, 
        const char *colname) {
    sqlite3_stmt *stmt;
    sprintf(g_sql_buf[2], "create table [%s].[%s %s] (level int, label text, "
            "primary key(level), unique(label));", iname, factor_type, colname);
    int res;
    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
    if (res == SQLITE_OK) sqlite3_step(stmt);
    sqlite3_finalize(stmt);
    return res; /* error on dup name? */
}
 
/* copy a factor|ordered table from a sdf(db) to another sdf(db) */
int _copy_factor_levels2(const char *factor_type, const char *iname_src,
        const char *colname_src, const char *iname_dst, const char *colname_dst) {
    sqlite3_stmt *stmt;
    int res;
    res = _create_factor_table2(iname_dst, factor_type, colname_dst);
    if (res == SQLITE_OK) {
        sprintf(g_sql_buf[2], "insert into [%s].[%s %s] select * from [%s].[%s %s]",
                iname_dst, factor_type, colname_dst, iname_src, factor_type,
                colname_src);
        res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);
        if (res == SQLITE_OK) res = sqlite3_step(stmt);
        sqlite3_finalize(stmt);
    }
    return res; /* error on dup name? */
}
 
 
/* assumes stmt has col_cnt + 1 columns, col 0 is [row name]. returned SEXP is 
 * not UNPROTECT-ed, user will have to do that. will create the names &
 * attach factor infos */
static SEXP _setup_df_sexp1(sqlite3_stmt *stmt, const char *iname, 
        int col_cnt, int row_cnt, int *dup_indices) {
    SEXP ret, names, value, class = R_NilValue;
    int i, type;
    const char *colname, *coltype;
 
    PROTECT(ret = NEW_LIST(col_cnt));
 
    /* set up names. */
    PROTECT(names = NEW_CHARACTER(col_cnt));
    for (i = 0; i < col_cnt; i++) {
        colname = sqlite3_column_name(stmt, i+1);  /* +1 bec [row name is 0] */
        coltype = sqlite3_column_decltype(stmt, i+1);
 
        if (dup_indices[i] == 0) {
            strcpy(g_sql_buf[1], colname);
        } else {
            sprintf(g_sql_buf[1], "%s.%d", colname, dup_indices[i]);
        }
 
        SET_STRING_ELT(names, i, mkChar(g_sql_buf[1]));
 
        if (strcmp(coltype, "text") == 0) {
            PROTECT(value = NEW_CHARACTER(row_cnt));
        } else if (strcmp(coltype, "double") == 0) {
            PROTECT(value = NEW_NUMERIC(row_cnt));
        } else if (strcmp(coltype, "bit") == 0) {
            PROTECT(value = NEW_LOGICAL(row_cnt));
        } else if (strcmp(coltype, "integer") == 0 || 
                   strcmp(coltype, "int") == 0) {
            PROTECT(value = NEW_INTEGER(row_cnt));
            type = _get_factor_levels1(iname, colname, value);
            if (type == VAR_FACTOR) {
                PROTECT(class = mkString("factor"));
                SET_CLASS(value, class);
                UNPROTECT(1);
            } else if (type == VAR_ORDERED) {
                PROTECT(class = NEW_CHARACTER(2));
                SET_STRING_ELT(class, 0, mkChar("ordered"));
                SET_STRING_ELT(class, 1, mkChar("factor"));
                SET_CLASS(value, class);
                UNPROTECT(1);
            }
        } else {
            Rprintf("Error: not supported type %s for %s\n", coltype, colname);
            UNPROTECT(2); /* unprotect ret, names */
            return R_NilValue;
        }
 
        SET_VECTOR_ELT(ret, i, value);
        UNPROTECT(1); /* unprotect value */
                    
    }
    SET_NAMES(ret, names);
    UNPROTECT(1); /* unprotect names only */
    return ret;
}
 
/* expected that 1st col of stmt is [row name], so we start with index 1 */
static int _add_row_to_df(SEXP df, sqlite3_stmt *stmt, int row, int ncols) {
    SEXP vec;
    int type = -1, is_null, i;
    for (i = 1; i <= ncols; i++) {
        vec = VECTOR_ELT(df, i-1);
        type = TYPEOF(vec);
 
        is_null = (sqlite3_column_type(stmt, row) == SQLITE_NULL);
 
        if (type == STRSXP || type == CHARSXP) {
            SET_STRING_ELT(vec, row, mkChar((char *)sqlite3_column_text(stmt, i)));
        } else if (type == INTSXP) {
            INTEGER(vec)[row] = sqlite3_column_int(stmt, i);
        } else if (type == REALSXP) {
            REAL(vec)[row] = sqlite3_column_double(stmt, i);
        } else if (type == LGLSXP) {
            LOGICAL(vec)[row] = sqlite3_column_int(stmt, i);
        }
    }
    return type;
}
 
/* set ordinary df row name as numbers */
static void _set_rownames2(SEXP df) {
    SEXP value = VECTOR_ELT(df, 0);
    int len = LENGTH(value), i;
    PROTECT(value = NEW_CHARACTER(len));
    for (i = 0; i < len; i++) {
        sprintf(g_sql_buf[2], "%d", i+1);
        SET_STRING_ELT(value, i, mkChar(g_sql_buf[2]));
    }
 
    SET_ROWNAMES(df, value);
    UNPROTECT(1);
}
 
 
/****************************************************************************
 * SDF FUNCTIONS
 ****************************************************************************/
 
SEXP sdf_create_sdf(SEXP df, SEXP name) {
    SEXP ret;
    char *iname; 
    int namelen, res, i, j;
 
    /* find free name, attach sdf, create sdf_attributes */
    iname = _create_sdf_skeleton1(name, &namelen, FALSE);
    
    if (iname != NULL) {
        int sql_len, sql_len2;
        sqlite3_stmt *stmt;
 
        /* create sdf_data table */
        SEXP names = GET_NAMES(df), variable, levels, var_class;
        int ncols = GET_LENGTH(names), type, *types;
        char *col_name, *class, *factor;
 
        /* variables for adding rownames */
        SEXP rownames;
        int nrows;
        char *row_name;
 
        /* TODO: put constraints on table after inserting everything? */
 
 
        /* 
         * create the create table and insert sql scripts by looping through
         * the columns of df
         */
        types = (int *)R_alloc(ncols, sizeof(int));
        sql_len = sprintf(g_sql_buf[0], "create table [%s].sdf_data ([row name] text", iname);
        sql_len2 = sprintf(g_sql_buf[1], "insert into [%s].sdf_data values(?", iname);
 
        for (i = 0; i < ncols; i++) {
            col_name = CHAR(STRING_ELT(names, i));
 
            /* add column definition to the create table sql */
            _expand_buf(0, sql_len+strlen(col_name)+10);
 
            variable = _getListElement(df, col_name);
            var_class = GET_CLASS(variable);
            class = (var_class == R_NilValue) ? NULL : CHAR(STRING_ELT(var_class, 0));
            type = TYPEOF(variable);
            types[i] = type;
 
            sql_len += sprintf(g_sql_buf[0]+sql_len, ", [%s] %s", col_name, 
                    _get_column_type(class, type));
 
            /* add handler to insert table sql */
            _expand_buf(1, sql_len+5);
            strcpy(g_sql_buf[1]+sql_len2, ",?");
            sql_len2 += 2; 
 
            /* create separate table for factors decode */
            if (class != NULL && (strcmp(class, "factor") == 0 || strcmp(class, "ordered") == 0)){
                if (_create_factor_table2(iname, class, col_name)) 
                    return R_NilValue; /* dup tbl name? */
 
                _sqlite_exec("begin");
                levels = GET_LEVELS(variable);
                sprintf(g_sql_buf[2], "insert into [%s].[%s %s] values(?, ?);",
                        iname, class, col_name);
                res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, NULL);
                if (_sqlite_error(res)) return R_NilValue; /* dup tbl name? */
 
                for (j = 0; j < GET_LENGTH(levels); j++) {
                    sqlite3_reset(stmt);
                    factor = CHAR(STRING_ELT(levels, j));
                    sqlite3_bind_int(stmt, 1, j+1);
                    sqlite3_bind_text(stmt, 2, factor, -1, SQLITE_STATIC);
                    sqlite3_step(stmt);
                }
                sqlite3_finalize(stmt);
                _sqlite_exec("commit");
            }
        }
        
        _expand_buf(0,sql_len+35);
        sql_len += sprintf(g_sql_buf[0]+sql_len, ", primary key([row name]));");
        /* sql_len += sprintf(g_sql_buf[0]+sql_len, ");"); */
        res = _sqlite_exec(g_sql_buf[0]);
        if (_sqlite_error(res)) return R_NilValue; /* why? */
 
        /*
         * add the data in df to the sdf
         */
        rownames = getAttrib(df, R_RowNamesSymbol);
        nrows = GET_LENGTH(rownames);
 
        _sqlite_error(_sqlite_exec("begin"));
        sprintf(g_sql_buf[1]+sql_len2, ")");
        res = sqlite3_prepare(g_workspace, g_sql_buf[1], -1, &stmt, NULL);
 
        for (i = 0; i < nrows; i++) {
            sqlite3_reset(stmt);
 
            /* since this is coming from a real dataframe, we're assured that
             * the rownames are unique */
            if (IS_CHARACTER(rownames)) {
                row_name = CHAR(STRING_ELT(rownames, i));
                if (*row_name) /* if not empty string */
                    sqlite3_bind_text(stmt, 1, row_name, strlen(row_name), SQLITE_STATIC);
                else sqlite3_bind_int(stmt, 1, i);
            } else if (IS_INTEGER(rownames)) {
                sqlite3_bind_int(stmt, 1, INTEGER(rownames)[i]);
            } else sqlite3_bind_int(stmt, 1, i);
 
            for (j = 0; j < ncols; j++) {
                variable = VECTOR_ELT(df, j);
                switch(types[j]) {
                    case INTSXP : 
                        sqlite3_bind_int(stmt, j+2, INTEGER(variable)[i]);
                        break;
                    case REALSXP:
                        sqlite3_bind_double(stmt, j+2, REAL(variable)[i]);
                        break;
                    case CHARSXP:
                    case STRSXP:
                        sqlite3_bind_text(stmt, j+2, CHAR_ELT(variable, i), -1, SQLITE_STATIC);
                }
                /* TODO: handle NA's & NULL's */
            }
 
            res = sqlite3_step(stmt);
            if (res != SQLITE_DONE) { 
                _sqlite_exec("ROLLBACK");
                sqlite3_finalize(stmt);
                Rprintf("SQLITE ERROR: %s\n", sqlite3_errmsg(g_workspace));
                return R_NilValue; /* why? */
            }
        }
                
        sqlite3_finalize(stmt);
        _sqlite_error(_sqlite_exec("COMMIT"));
 
        /* create a new object representing the sdf */
        ret = _create_sdf_sexp(iname);
 
    } else {
        Rprintf("ERROR: unable to create a sqlite data frame.\n");
        ret = R_NilValue;
    }
        
    return ret;
}
 
SEXP sdf_get_names(SEXP sdf) {
    char *iname;
    iname = SDF_INAME(sdf);
    if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;
 
    int len;
    len = sprintf(g_sql_buf[0], "select * from [%s].sdf_data;", iname);
 
    sqlite3_stmt *stmt;
    int res, i;
   
    res = sqlite3_prepare(g_workspace, g_sql_buf[0], len, &stmt, NULL);
    if (_sqlite_error(res)) return R_NilValue;
 
    SEXP ret;
 
    len = sqlite3_column_count(stmt)-1;
    PROTECT(ret = NEW_CHARACTER(len));
 
    for (i = 0; i < len; i++) {
        SET_STRING_ELT(ret, i, mkChar(sqlite3_column_name(stmt, i+1)));
    }
 
    sqlite3_finalize(stmt);
    UNPROTECT(1);
    return ret;
}
 
SEXP sdf_get_length(SEXP sdf) {
    char *iname;
    int len;
    sqlite3_stmt *stmt;
    int res;
 
    iname  = CHAR(STRING_ELT(_getListElement(sdf, "iname"),0));
    sprintf(g_sql_buf[0], "select * from [%s].sdf_data;", iname);
    if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;
 
    res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, NULL);
    if (_sqlite_error(res)) return R_NilValue;
 
    len = sqlite3_column_count(stmt)-1;
    sqlite3_finalize(stmt);
    return ScalarInteger(len);
}
 
/* get row count */
SEXP sdf_get_row_count(SEXP sdf) {
    char *iname = CHAR(STRING_ELT(_getListElement(sdf, "iname"),0));
    if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;
 
    int nrow;
    SEXP ret;
   
    nrow = _get_row_count2(iname, TRUE);
 
    if (nrow < 1) ret = R_NilValue;
    else ret = ScalarInteger(nrow);
 
    return ret;
}
 
    
SEXP sdf_import_table(SEXP _filename, SEXP _name, SEXP _sep, SEXP _quote, 
        SEXP _rownames, SEXP _colnames) {
    char *filename = CHAR_ELT(_filename, 0);
    FILE *f = fopen(filename, "r");
 
    if (f == NULL) {
        Rprintf("Error: File %s does not exist.", filename);
        return R_NilValue;
    }
 
    /*char *sep = CHAR_ELT(_sep, 0), *quote = CHAR_ELT(_quote,0);
    char *name = CHAR_ELT(_name, 0);*/
 
    /* create the table */
    /* insert the stuffs */
    /* register to workspace */
 
    return R_NilValue;
}
 
SEXP sdf_get_index(SEXP sdf, SEXP row, SEXP col, SEXP new_sdf) {
    SEXP ret = R_NilValue;
    char *iname = SDF_INAME(sdf);
    sqlite3_stmt *stmt;
    int buflen = 0, idxlen, col_cnt, row_cnt, index;
    int i, j,res;
    int *col_indices, col_index_len, *dup_indices;
    int row_index_len = 0, force_new_df = LOGICAL(new_sdf)[0];
    const char *colname;
 
    /* dup_indices is used to handle when same column chosen more than once.
     * e.g. iris[,c(1,1)] have names Sepal.Length, Sepal.Length.1 *
     * col_indices is used to store the column index of selected columns
     * in the sdf_data table */
 
    if (!USE_SDF1(iname, TRUE, TRUE)) return R_NilValue;
 
    sprintf(g_sql_buf[0], "select * from [%s].sdf_data ", iname);
    res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);
    if (_sqlite_error(res)) return R_NilValue;
 
    buflen = sprintf(g_sql_buf[0], "select [row name]");
    idxlen = LENGTH(col);
    col_cnt = sqlite3_column_count(stmt) - 1;
 
    /*
     * PROCESS COLUMN INDEX: set columns in the select statement
     */
    if (col == R_NilValue) {
        for (i = 1; i < col_cnt+1; i++) {
            colname = sqlite3_column_name(stmt, i);
            buflen += sprintf(g_sql_buf[0]+buflen, ",[%s]", colname);
        }
        _expand_buf(0, buflen+20+strlen(iname));
        buflen += sprintf(g_sql_buf[0]+buflen, " from [%s].sdf_data", iname);
        col_index_len = col_cnt;
        col_indices = (int *)R_alloc(col_cnt, sizeof(int));
        dup_indices = (int *)R_alloc(col_cnt, sizeof(int));
        for (i = 0; i < col_cnt; i++) { col_indices[i] = i; dup_indices[i] = 0; }
    } else if (col == R_NilValue || idxlen < 1) {
        sqlite3_finalize(stmt);
        return R_NilValue;
    } else if (IS_NUMERIC(col)) {
        col_indices = (int *)R_alloc(idxlen, sizeof(int));
        dup_indices = (int *)R_alloc(idxlen, sizeof(int));
        col_index_len = 0;
 
        for (i = 0; i < idxlen; i++) {
            /* no need to correct for 0 base because col 0 is [row name] */
            index = ((int) REAL(col)[i]); 
            if (index > col_cnt) {
                sqlite3_finalize(stmt);
                error("undefined columns selected\n");
            } else if (index > 0) {
                colname = sqlite3_column_name(stmt,index);
                buflen += sprintf(g_sql_buf[0]+buflen, ",[%s]", colname);
                dup_indices[col_index_len] = 0;
                for (j = 0; j < col_index_len; j++) {
                    if (col_indices[j] == index) dup_indices[col_index_len]++;
                }
                col_indices[col_index_len++] = index;
            } else if (index < 0) {
                sqlite3_finalize(stmt);
                error("negative indices not supported.\n");
            }
        }
 
        if (col_index_len == 0) {
            sqlite3_finalize(stmt);
            Rprintf("Error: no indices detected??\n");
            return R_NilValue;
        } else { 
            _expand_buf(0, buflen+20+strlen(iname));
            buflen += sprintf(g_sql_buf[0]+buflen, " from [%s].sdf_data", iname);
        }
    } else if (IS_INTEGER(col)) {
        /* identical logic with IS_NUMERIC, except that we don't have to cast
         * stuffs from SEXP col. the alternative is doing if idxlen times. */
        col_indices = (int *)R_alloc(idxlen, sizeof(int));
        dup_indices = (int *)R_alloc(idxlen, sizeof(int));
        col_index_len = 0;
 
        for (i = 0; i < idxlen; i++) {
            /* no need to correct for 0 base because col 0 is [row name] */
            index = INTEGER(col)[i];
            if (index > col_cnt) {
                sqlite3_finalize(stmt);
                error("undefined columns selected\n");
            } else if (index > 0) {
                colname = sqlite3_column_name(stmt,index);
                buflen += sprintf(g_sql_buf[0]+buflen, ",[%s]", colname);
                dup_indices[col_index_len] = 0;
                for (j = 0; j < col_index_len; j++) {
                    if (col_indices[j] == index) dup_indices[col_index_len]++;
                }
                col_indices[col_index_len++] = index;
            } else if (index < 0) {
                sqlite3_finalize(stmt);
                error("negative indices not supported.\n");
            }
        }
 
        if (col_index_len == 0) {
            sqlite3_finalize(stmt);
            Rprintf("Error: no indices detected??\n");
            return R_NilValue;
        } else { 
            _expand_buf(0, buflen+20+strlen(iname));
            buflen += sprintf(g_sql_buf[0]+buflen, " from [%s].sdf_data", iname);
        }
    } else if (IS_LOGICAL(col)) {
        /* recycling stuff, so max column output is the # of cols in the df */
        col_indices = (int *)R_alloc(col_cnt, sizeof(int));
        dup_indices = (int *)R_alloc(col_cnt, sizeof(int));
        col_index_len = 0;
        for (i = 0; i < col_cnt; i++) {
            if (LOGICAL(col)[i%idxlen]) {
                buflen += sprintf(g_sql_buf[0]+buflen, ",[%s]", 
                        sqlite3_column_name(stmt,i+1));
 
                dup_indices[col_index_len] = 0;
                col_indices[col_index_len++] = i;
            }
 
        }
 
        if (col_index_len == 0) {
            sqlite3_finalize(stmt);
            warning("no column selected.\n");
            return R_NilValue;
        } else { 
            _expand_buf(0, buflen+20+strlen(iname));
            buflen += sprintf(g_sql_buf[0]+buflen, " from [%s].sdf_data", iname);
        }
    } else {
        sqlite3_finalize(stmt);
        error("don't know how to handle column index.\n");
    }
 
    /* 
     * PROCESS ROW INDEX: setup limit or where clause
     */
    idxlen = LENGTH(row);
    if (row == R_NilValue) {
        if (col_index_len == 1 && !force_new_df) {
            ret = sdf_get_variable(sdf, mkString(sqlite3_column_name(stmt,col_indices[0])));
            sqlite3_finalize(stmt);
            return ret;
        } if (col_index_len > 1 || (force_new_df && col_index_len == 1)) {
            /* create a new SDF, logic similar to sdf_create_sdf */
 
            /* find a new name. data<n> ? */
            char *iname2;
            int namelen, sql_len, sql_len2;
 
            iname2 = _create_sdf_skeleton1(R_NilValue, &namelen, TRUE);
 
            /* create sdf_data table */
            sql_len = sprintf(g_sql_buf[1], "create table [%s].sdf_data ([row name] text", iname2);
            sql_len2 = 0;
 
            for (i = 0; i < col_index_len; i++) {
                colname = sqlite3_column_name(stmt, col_indices[i]);
                if (dup_indices[i] == 0) {
                    sql_len += sprintf(g_sql_buf[1]+sql_len, ", [%s] %s", colname,
                            sqlite3_column_decltype(stmt, col_indices[i]));
                } else {
                    /* un-duplicate col names by appending num, just like R */
                    sql_len += sprintf(g_sql_buf[1]+sql_len, ", [%s.%d] %s", colname,
                            dup_indices[i], 
                            sqlite3_column_decltype(stmt, col_indices[i]));
                }
 
                /* deal with possibly factor columns */
                if (_is_factor2(iname, "factor", colname)) {
                    /* found a factor column */
 
                    if (dup_indices[i] == 0)  {
                        _copy_factor_levels2("factor", iname, colname, iname2, colname);
                    } else {
                        sprintf(g_sql_buf[3], "%s.%d", colname, dup_indices[i]);
                        _copy_factor_levels2("factor", iname, colname, iname2, 
                                g_sql_buf[3]);
                    }
                }
 
                /* and deal with ordered factors too... life is hard, then you die */
                if (_is_factor2(iname, "ordered", colname)) {
                    
                    if (dup_indices[i] == 0)  {
                        _copy_factor_levels2("ordered", iname, colname, iname2, 
                                colname);
                    } else {
                        sprintf(g_sql_buf[3], "%s.%d", colname, dup_indices[i]);
                        _copy_factor_levels2("ordered", iname, colname, iname2, 
                                g_sql_buf[3]);
                    }
                }
                                
            }
 
            /* don't need it anymore */
            sqlite3_finalize(stmt);
 
            /* no table index created, that's for v2 I hope*/
            sql_len += sprintf(g_sql_buf[1]+sql_len, ");");
            res = _sqlite_exec(g_sql_buf[1]);
            if (_sqlite_error(res)) { Rprintf("here? %s\n", g_sql_buf[1]); return R_NilValue; }
 
            /* insert data (with row names). buf[0] contains a select */
            sprintf(g_sql_buf[1], "insert into [%s].sdf_data %s", iname2,
                    g_sql_buf[0]);
            res = _sqlite_exec(g_sql_buf[1]);
            if (_sqlite_error(res)) return R_NilValue;
 
            /* remove protection */
            UNUSE_SDF2(iname2);
 
            /* create SEXP for the SDF */
            ret = _create_sdf_sexp(iname2);
            return ret;
        } else { sqlite3_finalize(stmt); return R_NilValue; }
    } else if (row == R_NilValue || idxlen < 1) {
        sqlite3_finalize(stmt);
        return R_NilValue;
    } else if (IS_NUMERIC(row)) {
        sqlite3_finalize(stmt);
 
        sprintf(g_sql_buf[1], "[%s].sdf_data", iname);
        row_cnt = _get_row_count2(g_sql_buf[1], 0);
 
        /* append " limit ?,1" to the formed select statement */
        _expand_buf(0, buflen+10);
        buflen += sprintf(g_sql_buf[0]+buflen, " limit ?,1");
        /*Rprintf("sql [%d]: %s\n", buflen, g_sql_buf[0]); */
 
        index = ((int) REAL(row)[0]) - 1;
        if (idxlen < 0 && idxlen == 1) return R_NilValue;
 
        res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);
        if (_sqlite_error(res)) return R_NilValue;
 
        sqlite3_bind_int(stmt, 1, index);
        res = sqlite3_step(stmt);
 
        /* create data frame */
        ret = _setup_df_sexp1(stmt, iname, col_index_len, idxlen, dup_indices);
        if (ret == R_NilValue) { sqlite3_finalize(stmt); return R_NilValue; }
 
        /* put data in it */
        if (index >= 0 && index < row_cnt) {
            _add_row_to_df(ret, stmt, row_index_len++, col_index_len);
        }
 
        for (i = 1; i < idxlen; i++) {
            sqlite3_reset(stmt);
            index = ((int) REAL(row)[i]) - 1;
            if (index >= 0 && index < row_cnt) {
                sqlite3_bind_int(stmt, 1, index);
                res = sqlite3_step(stmt);
                _add_row_to_df(ret, stmt, row_index_len++, col_index_len);
            }
        }
        sqlite3_finalize(stmt);
 
        /* shrink vectors */
        if (row_index_len < idxlen) {
            for (i = 0; i < col_index_len; i++) {
                SET_VECTOR_ELT(ret, i, 
                        _shrink_vector(VECTOR_ELT(ret, i), row_index_len));
            }
        }
 
        UNPROTECT(1); /* for the ret from _setup_df_sexp1 */
    } else if (IS_INTEGER(row)) {
        /* same stuff as with IS_INTEGER, except for setting of var index*/
        sqlite3_finalize(stmt);
 
        sprintf(g_sql_buf[1], "[%s].sdf_data", iname);
        row_cnt = _get_row_count2(g_sql_buf[1], 0);
 
        /* append " limit ?,1" to the formed select statement */
        _expand_buf(0, buflen+10);
        buflen += sprintf(g_sql_buf[0]+buflen, " limit ?,1");
 
        index = INTEGER(row)[0] - 1;
        if (idxlen < 0 && idxlen == 1) return R_NilValue;
 
        res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);
        if (_sqlite_error(res)) return R_NilValue;
 
        sqlite3_bind_int(stmt, 1, index);
        res = sqlite3_step(stmt);
 
        /* create data frame */
        ret = _setup_df_sexp1(stmt, iname, col_index_len, idxlen, dup_indices);
        if (ret == R_NilValue) { sqlite3_finalize(stmt); return R_NilValue; }
 
        /* put data in it */
        if (index >= 0 && index < row_cnt) {
            _add_row_to_df(ret, stmt, row_index_len++, col_index_len);
        }
 
        for (i = 1; i < idxlen; i++) {
            sqlite3_reset(stmt);
            index = INTEGER(row)[i] - 1;
            if (index >= 0 && index < row_cnt) {
                sqlite3_bind_int(stmt, 1, index);
                res = sqlite3_step(stmt); 
                _add_row_to_df(ret, stmt, row_index_len++, col_index_len);
            }
        }
        sqlite3_finalize(stmt);
 
        /* shrink vectors */
        if (row_index_len < idxlen) {
            for (i = 0; i < col_index_len; i++) {
                SET_VECTOR_ELT(ret, i, 
                        _shrink_vector(VECTOR_ELT(ret, i), row_index_len));
            }
        }
        UNPROTECT(1); /* for the ret from _setup_df_sexp1 */
    } else if (IS_LOGICAL(row)) {
        /* */
        sqlite3_finalize(stmt);
        int est_row_cnt = 0;
 
        sprintf(g_sql_buf[1], "[%s].sdf_data", iname);
        row_cnt = _get_row_count2(g_sql_buf[1], 0);
 
        /* append " limit ?,1" to the formed select statement */
        _expand_buf(0, buflen+10);
        buflen += sprintf(g_sql_buf[0]+buflen, " limit ?,1");
 
        /* find if there is any TRUE element in the vector */
        for (i = 0; i < idxlen && i < row_cnt; i++) {
            if (LOGICAL(row)[i]) break;
        }
 
        if (i < idxlen && i < row_cnt) {
            est_row_cnt = (idxlen-i) * (row_cnt/idxlen);
            if (row_cnt%idxlen > i) est_row_cnt += (row_cnt%idxlen - i);
 
            res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, 0);
            if (_sqlite_error(res)) return R_NilValue;
 
            sqlite3_bind_int(stmt, 1, i);
            sqlite3_step(stmt);
            ret = _setup_df_sexp1(stmt, iname, col_index_len, est_row_cnt, dup_indices);
            _add_row_to_df(ret, stmt, row_index_len++, col_index_len);
            
            for (i++ ; i < row_cnt; i++) {
                if (LOGICAL(row)[i%idxlen]) {
                    sqlite3_reset(stmt);
                    sqlite3_bind_int(stmt, 1, i);
                    sqlite3_step(stmt);
                    _add_row_to_df(ret, stmt, row_index_len++, col_index_len);
                }
            }
        }
 
        sqlite3_finalize(stmt);
 
        /* shrink vectors */
        if (est_row_cnt > 0 && row_index_len < est_row_cnt) {
            for (i = 0; i < col_index_len; i++) {
                SET_VECTOR_ELT(ret, i, 
                        _shrink_vector(VECTOR_ELT(ret, i), row_index_len));
            }
        }
 
        UNPROTECT(1); /* for the ret from _setup_df_sexp1 */
    }
 
    UNUSE_SDF2(iname);
 
    if (ret != R_NilValue && col_index_len > 1) {
        SET_CLASS(ret, mkString("data.frame"));
        _set_rownames2(ret);
    } else if (ret != R_NilValue && col_index_len == 1) {
        ret = VECTOR_ELT(ret, 0);
    }
 
    return ret;
}
 
SEXP sdf_rbind(SEXP sdf, SEXP data) {
    char *class;
    char *iname = SDF_INAME(sdf), *colname;
    SEXP ret = R_NilValue, col, names, levels, rownames;
    int ncols, buflen, buflen2, i, j, res, nrows, *types, rownames_type;
    sqlite3_stmt *stmt;
    const char *type;
    char *rn_str; 
 
    class = CHAR_ELT(GET_CLASS(data), 0);
    if (strcmp(class,"data.frame") == 0) {
        /* check table names, types. variables may not be in same order, as
         * long as names and types are identical. */
        names = GET_NAMES(data);
        ncols = GET_LENGTH(names);
 
        buflen = sprintf(g_sql_buf[0], "select 1");
        for (i = 0; i < ncols; i++) {
            _expand_buf(0, buflen + 100);
            buflen += sprintf(g_sql_buf[0]+buflen, ",[%s]", CHAR_ELT(names, i));
        }
        _expand_buf(0, buflen + 100);
        sprintf(g_sql_buf[0]+buflen, " from [%s].sdf_data", iname);
 
        res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, NULL);
        if (res != SQLITE_OK) { /* column name mismatch */
            _sqlite_error(res);
            Rprintf("Error: column names should exactly match the names of the SDF.\n");
            return ret;
        }
 
        /* now check the type */
        types = (int *)R_alloc(ncols, sizeof(int));
        for (i = 0; i< ncols; i++) {
            type = sqlite3_column_decltype(stmt, i+1); /* +1 bec 1st col is "1" */
 
            col = VECTOR_ELT(data, i);
            /* class = CHAR_ELT(GET_CLASS(col),0); */
 
            if (strcmp(type, "double") == 0 && IS_NUMERIC(col)) types[i] = SQLITE_FLOAT;
            else if (strcmp(type, "text") == 0 && IS_CHARACTER(col)) types[i] = SQLITE_TEXT;
            else if (strcmp(type, "int") == 0) { 
                types[i] = SQLITE_INTEGER;
                if (isUnordered(col)) {  /* test if factor */
                    colname = CHAR_ELT(names, i);
                    sqlite3_stmt *stmt2;
                    if (!_is_factor2(iname, "factor", colname)) break;
                    sprintf(g_sql_buf[1], "[%s].[factor %s]", iname, colname);
                    nrows = _get_row_count2(g_sql_buf[1], 0);
 
                    /* levels in sdf must be >= levels in df */
                    levels = GET_LEVELS(col);
                    if (nrows < GET_LENGTH(levels)) {
                        Rprintf("Error: The data frame variable %s has more levels than"
                                " its SDF counterpart.\n", colname);
                        break;
                    }
 
                    sprintf(g_sql_buf[2], "select label from %s order by level", g_sql_buf[1]);
                    res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt2, NULL);
                    _sqlite_error(res);
                    
                    for (j = 0; j < GET_LENGTH(levels); j++)  {
                        sqlite3_step(stmt2);
                        if (strcmp((char *)sqlite3_column_text(stmt2, 0), CHAR_ELT(levels, j)) != 0) {
                            Rprintf("Error: Level label mismatch in variable %s, [%s] in data frame"
                                    " vs [%s] in SDF.\n", colname, (char *)sqlite3_column_text(stmt2, 0),
                                    CHAR_ELT(levels, j));
                            sqlite3_finalize(stmt2);
                            break;
                        }
                    }
                    sqlite3_finalize(stmt2);
                } else if (isOrdered(col)) { /* same with ordered */
                    colname = CHAR_ELT(names, i);
                    sqlite3_stmt *stmt2;
                    if (!_is_factor2(iname, "ordered", colname)) break;
                    sprintf(g_sql_buf[1], "[%s].[ordered %s]", iname, colname);
                    nrows = _get_row_count2(g_sql_buf[1], 0);
 
                    /* levels in sdf must be >= levels in df */
                    levels = GET_LEVELS(col);
                    if (nrows < GET_LENGTH(levels)) {
                        Rprintf("Error: The data frame variable %s has more levels than"
                                " its SDF counterpart.\n", colname);
                        break;
                    }
 
                    sprintf(g_sql_buf[2], "select label from %s order by level", g_sql_buf[1]);
                    sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt2, NULL);
                    
                    for (j = 0; j < GET_LENGTH(levels); j++)  {
                        sqlite3_step(stmt2);
                        if (strcmp((char *)sqlite3_column_text(stmt2, 0), CHAR_ELT(levels, j)) != 0) {
                            Rprintf("Error: Level label mismatch in variable %s, [%s] in data frame"
                                    " vs [%s] in SDF.\n", colname, (char *)sqlite3_column_text(stmt2, 0),
                                    CHAR_ELT(levels, j));
                            sqlite3_finalize(stmt2);
                            break;
                        }
                    }
                    sqlite3_finalize(stmt2);
                }  /* else if class == "ordered" */
            } /* if integer */
        } /* for loop for column type checking */
 
        sqlite3_finalize(stmt);
        if (i < ncols) {
            Rprintf("Error: type mismatch at variable %s\n", CHAR_ELT(names, i));
            return ret;
        }
 
        /* create the insert statement */
        buflen = sprintf(g_sql_buf[0], "insert into [%s].sdf_data([row name]", iname);
        buflen2 = sprintf(g_sql_buf[1], "values(?");
        for (i = 0; i < ncols; i++) {
            _expand_buf(0, buflen+100);
            buflen += sprintf(g_sql_buf[0]+buflen, ",[%s]", CHAR_ELT(names, i));
            _expand_buf(1, buflen2+5);
            buflen2 += sprintf(g_sql_buf[1]+buflen2, ",?");
        }
 
        _expand_buf(2, buflen + buflen2 + 10);
        sprintf(g_sql_buf[2], "%s) %s)", g_sql_buf[0], g_sql_buf[1]);
        res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, NULL);
        if (_sqlite_error(res)) return ret;
 
        /* finally, add rows */
        rownames = getAttrib(data, R_RowNamesSymbol); /* GET_ROWNAMES(data); */
        nrows = LENGTH(rownames);
        rownames_type = TYPEOF(rownames);
 
        _sqlite_begin;
        for (i = 0; i < nrows; i++) {
            sqlite3_reset(stmt);
            if (rownames_type == STRSXP) {
                rn_str = CHAR_ELT(rownames, i);
            } else if (rownames_type == INTSXP) {
                sprintf(g_sql_buf[1], "%d", INTEGER(rownames)[i]);
                rn_str = g_sql_buf[1];
            } else if (rownames_type == REALSXP) {
                sprintf(g_sql_buf[1], "%f", REAL(rownames)[i]);
                rn_str = g_sql_buf[1];
            } else {
                sprintf(g_sql_buf[1], "%d", i);
                rn_str = g_sql_buf[1];
            }
            sqlite3_bind_text(stmt, 1, rn_str, -1, SQLITE_STATIC);
 
            for (j = 0; j < ncols; j++) {
                col = VECTOR_ELT(data, j);
                switch(types[j]) {
                    case SQLITE_FLOAT:
                        sqlite3_bind_double(stmt, j+2, REAL(col)[i]); break;
                    case SQLITE_TEXT:
                        sqlite3_bind_text(stmt, j+2, CHAR_ELT(col, i), -1, SQLITE_STATIC); break;
                    case SQLITE_INTEGER:
                    default:  /* runtime error if we're wrong */
                        sqlite3_bind_int(stmt, j+2, INTEGER(col)[i]); break;
                }
            }
 
            res = sqlite3_step(stmt);
            if (res != SQLITE_DONE) { /* row name pk conflict */
                for (j = 1; res != SQLITE_DONE; j++) { /* _normally_, j shouldn't overflow */
                    sqlite3_reset(stmt);
                    sprintf(g_sql_buf[2], "%s-%d", rn_str, j);
                    sqlite3_bind_text(stmt, 1, g_sql_buf[2], -1, SQLITE_STATIC);
                    res = sqlite3_step(stmt);
                }
            } 
        }
 
        sqlite3_finalize(stmt);
        _sqlite_commit;
        ret = sdf;
    } else if (strcmp(class,"sqlite.data.frame") == 0) {
    }
 
    return ret;
}
 
 
SEXP sdf_get_iname(SEXP sdf) {
    char *iname, *sql;
    sqlite3_stmt *stmt;
    SEXP ret, names;
 
    iname = SDF_INAME(sdf);
    if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;
 
    sql = "select internal_name, rel_filename from workspace where internal_name=?";
    sqlite3_prepare(g_workspace, sql, -1, &stmt, NULL);
    sqlite3_bind_text(stmt, 1, iname, -1, SQLITE_STATIC);
    sqlite3_step(stmt);
 
    PROTECT(ret = NEW_CHARACTER(2));
    SET_STRING_ELT(ret, 0, mkChar((char *)sqlite3_column_text(stmt, 0)));
    SET_STRING_ELT(ret, 1, mkChar((char *)sqlite3_column_text(stmt, 1)));
 
    PROTECT(names = NEW_CHARACTER(2));
    SET_STRING_ELT(names, 0, mkChar("Internal Name"));
    SET_STRING_ELT(names, 1, mkChar("File"));
 
    SET_NAMES(ret, names);
 
    sqlite3_finalize(stmt);
    UNPROTECT(2);
 
    return ret;
}

Rev 3473	Rev 3684
1	`#include <stdio.h>`	1	`#include <stdio.h>`
2	`#include <string.h>`	2	`#include <string.h>`
3	`#include "sqlite_dataframe.h"`	3	`#include "sqlite_dataframe.h"`
4		4
5	`/****************************************************************************`	5	`/****************************************************************************`
6	`* UTILITY FUNCTIONS`	6	`* UTILITY FUNCTIONS`
7	`****************************************************************************/`	7	`****************************************************************************/`
8		8
9	`/* if user supplied a name, return that name. otherwise, use the`	9	`/* if user supplied a name (1st arg), return that name. otherwise, use the`
10	`* default "data" */`	10	`* default "data". rname is R name, iname is internal name */`
11	`static int _check_sdf_name(SEXP name, char rname, char iname, int *file_idx) {`	11	`static int _check_sdf_name(SEXP name, char rname, char iname, int *file_idx) {`
12	`int namelen = 0;`	12	`int namelen = 0;`
13		13
14	`/* check if arg name is supplied */`	14	`/* check if arg name is supplied */`
15	`if (name == R_NilValue) {`	15	`if (name == R_NilValue) {`
16	`*rname = "data";`	16	`*rname = "data";`
17	`namelen = 5;`	17	`namelen = 5;`
18	`iname = (char)R_alloc(13, sizeof(char)); /* data10000.db\0 */`	18	`iname = (char)R_alloc(13, sizeof(char)); /* .SQLiteDF/data10000.db\0 */`
19	`*file_idx = 1;`	19	`*file_idx = 1;`
20	`sprintf(iname, "%s%d.db", rname, *file_idx);`	20	`sprintf(iname, "%s%d.db", rname, *file_idx);`
21	`} else if (IS_CHARACTER(name)) {`	21	`} else if (IS_CHARACTER(name)) {`
22	`*rname = CHAR_ELT(name,0);`	22	`*rname = CHAR_ELT(name,0);`
23	`if (!_is_r_sym(*rname)) {`	23	`if (!_is_r_sym(*rname)) {`
24	`Rprintf("Error: supplied name \"%s\"is not a valid R symbol.\n",`	24	`error("supplied name \"%s\"is not a valid R symbol.", *rname);`
25	`*rname);`	-
26	`return FALSE;`	-
27	`}`	25	`}`
28	`namelen = strlen(*rname);`	26	`namelen = strlen(*rname);`
29	`iname = (char)R_alloc(namelen + 9, sizeof(char)); /* <name>10000.db\0 */`	27	`iname = (char)R_alloc(namelen + 9, sizeof(char)); /* .SQLiteDF/<name>10000.db\0 */`
30	`sprintf(iname, "%s.db", rname);`	28	`sprintf(iname, "%s.db", rname);`
31	`} else {`	29	`} else {`
32	`Rprintf("Error: the supplied value for arg name is not a string.\n");`	30	`Rprintf("Error: the supplied value for arg name is not a string.\n");`
33	`}`	31	`}`
34	`return namelen;`	32	`return namelen;`
35	`}`	33	`}`
36		34
37	`static int _find_free_filename2(char rname, char iname, int namelen, int *file_idx) {`	35	`static int _find_free_filename2(char rname, char dirname, char *iname, int namelen, int *file_idx) {`
38	`sqlite3_stmt *stmt;`	36	`sqlite3_stmt *stmt;`
-		37	`char *tmp_iname;`
39	`sqlite3_prepare(g_workspace, "select 1 from workspace where internal_name=?", -1,`	38	`sqlite3_prepare(g_workspace, "select 1 from workspace where internal_name=?", -1,`
40	`&stmt, NULL);`	39	`&stmt, NULL);`
41	`do {`	40	`do {`
-		41	`if (dirname != NULL) {`
-		42	`sprintf(g_sql_buf[2], "%s/%s", dirname, *iname);`
-		43	`tmp_iname = g_sql_buf[2];`
-		44	`} else {`
-		45	`tmp_iname = *iname;`
-		46	`}`
-		47
42	`if (!_file_exists(*iname)) {`	48	`if (!_file_exists(tmp_iname)) {`
43	`sqlite3_reset(stmt);`	49	`sqlite3_reset(stmt);`
44	`sqlite3_bind_text(stmt, 1, *iname, -1, SQLITE_STATIC);`	50	`sqlite3_bind_text(stmt, 1, tmp_iname, -1, SQLITE_STATIC);`
45	`if (sqlite3_step(stmt) != SQLITE_ROW) break;`	51	`if (sqlite3_step(stmt) != SQLITE_ROW) break;`
46	`}`	52	`}`
47	`namelen = sprintf(iname, "%s%d.db", rname, ++(*file_idx)) - 3;`	53	`namelen = sprintf(iname, "%s%d.db", rname, ++(*file_idx)) - 3;`
48	`} while (*file_idx < 10000);`	54	`} while (*file_idx < 10000);`
49		55
50	`sqlite3_finalize(stmt);`	56	`sqlite3_finalize(stmt);`
51	`return *file_idx;`	57	`return *file_idx;`
52	`}`	58	`}`
53		59
54	`/* creates an sdf attribute table */`	60	`/* creates an sdf attribute table */`
55	`static int _create_sdf_attribute2(const char *iname) {`	61	`static int _create_sdf_attribute2(const char *iname) {`
56	`sprintf(g_sql_buf[2], "create table [%s].sdf_attributes(attr text, "`	62	`sprintf(g_sql_buf[2], "create table [%s].sdf_attributes(attr text, "`
57	`"value text, primary key (attr))", iname);`	63	`"value text, primary key (attr))", iname);`
58	`int res;`	64	`int res;`
59	`res = _sqlite_exec(g_sql_buf[2]);`	65	`res = _sqlite_exec(g_sql_buf[2]);`
60	`if (res == SQLITE_OK) {`	66	`if (res == SQLITE_OK) {`
61	`sprintf(g_sql_buf[2], "insert into [%s].sdf_attributes values ('name',"`	67	`sprintf(g_sql_buf[2], "insert into [%s].sdf_attributes values ('name',"`
62	`"'%s');", iname, iname);`	68	`"'%s');", iname, iname);`
63	`res = _sqlite_exec(g_sql_buf[2]);`	69	`res = _sqlite_exec(g_sql_buf[2]);`
64	`}`	70	`}`
65	`return res;`	71	`return res;`
66	`}`	72	`}`
67		73
68	`char _create_sdf_skeleton1(SEXP name, int onamelen, int protect) {`	74	`char _create_sdf_skeleton1(SEXP name, int onamelen, int protect) {`
69	`char iname, rname;`	75	`char iname, rname;`
70	`int namelen, file_idx = 0, res;`	76	`int namelen, file_idx = 0, res;`
71		77
72	`namelen = _check_sdf_name(name, &rname, &iname, &file_idx);`	78	`namelen = _check_sdf_name(name, &rname, &iname, &file_idx);`
73		79
74	`if (!namelen) return NULL;`	80	`if (!namelen) return NULL;`
75		81
76	`_find_free_filename2(rname, &iname, &namelen, &file_idx);`	82	`_find_free_filename2(rname, ".SQLiteDF", &iname, &namelen, &file_idx);`
77		83
78	`if (file_idx >= 10000) {`	84	`if (file_idx >= 10000) {`
79	`Rprintf("Error: cannot find free SDF name.\n");`	85	`Rprintf("Error: cannot find free SDF name.\n");`
80	`return NULL;`	86	`return NULL;`
81	`}`	87	`}`
82		88
83	`/* add to workspace */`	89	`/* add to workspace */`
84	`strcpy(g_sql_buf[3], iname);`	90	`sprintf(g_sql_buf[3], ".SQLiteDF/%s", iname);`
85	`iname[namelen] = 0; /* remove ".db" */`	91	`iname[namelen] = 0; /* remove ".db" */`
86	`res = _add_sdf1(g_sql_buf[3], iname);`	92	`res = _add_sdf1(g_sql_buf[3], iname);`
87	`if (_sqlite_error(res)) return NULL;`	93	`if (_sqlite_error(res)) return NULL;`
88		94
89	`/* detach SDF's if necessary to attach this one. if file does not`	95	`/* detach SDF's if necessary to attach this one. if file does not`
90	`* exist, then a sqlite db will be created after we make our 1st table */`	96	`* exist, then a sqlite db will be created after we make our 1st table */`
91	`if (!USE_SDF1(iname, FALSE, protect)) { /* _delete_sdf2(iname); */ return NULL; }`	97	`if (!USE_SDF1(iname, FALSE, protect)) { /* _delete_sdf2(iname); */ return NULL; }`
92		98
93		99
94	`/* create attributes table */`	100	`/* create attributes table */`
95	`res = _create_sdf_attribute2(iname);`	101	`res = _create_sdf_attribute2(iname);`
96	`if (_sqlite_error(res)) {`	102	`if (_sqlite_error(res)) {`
97	`sprintf(g_sql_buf[2], "detach '%s'", iname);`	103	`sprintf(g_sql_buf[2], "detach '%s'", iname);`
98	`_delete_sdf2(iname);`	104	`_delete_sdf2(iname);`
99	`return NULL;`	105	`return NULL;`
100	`}`	106	`}`
101		107
102	`if (onamelen) *onamelen = namelen;`	108	`if (onamelen) *onamelen = namelen;`
103	`return iname;`	109	`return iname;`
104	`}`	110	`}`
105	`/* checks if a column has a corresponding factor\|ordered table */`	111	`/* checks if a column has a corresponding factor\|ordered table */`
106	`int _is_factor2(const char iname, const char factor_type, const char *colname) {`	112	`int _is_factor2(const char iname, const char factor_type, const char *colname) {`
107	`sqlite3_stmt *stmt;`	113	`sqlite3_stmt *stmt;`
108	`sprintf(g_sql_buf[2], "select * from [%s].[%s %s]", iname, factor_type,`	114	`sprintf(g_sql_buf[2], "select * from [%s].[%s %s]", iname, factor_type,`
109	`colname);`	115	`colname);`
110	`int res;`	116	`int res;`
111	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);`	117	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);`
112	`sqlite3_finalize(stmt);`	118	`sqlite3_finalize(stmt);`
113	`return res == SQLITE_OK;`	119	`return res == SQLITE_OK;`
114	`}`	120	`}`
115		121
116	`/* create a factor\|ordered table */`	122	`/* create a factor\|ordered table */`
117	`int _create_factor_table2(const char iname, const char factor_type,`	123	`int _create_factor_table2(const char iname, const char factor_type,`
118	`const char *colname) {`	124	`const char *colname) {`
119	`sqlite3_stmt *stmt;`	125	`sqlite3_stmt *stmt;`
120	`sprintf(g_sql_buf[2], "create table [%s].[%s %s] (level int, label text, "`	126	`sprintf(g_sql_buf[2], "create table [%s].[%s %s] (level int, label text, "`
121	`"primary key(level), unique(label));", iname, factor_type, colname);`	127	`"primary key(level), unique(label));", iname, factor_type, colname);`
122	`int res;`	128	`int res;`
123	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);`	129	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);`
124	`if (res == SQLITE_OK) sqlite3_step(stmt);`	130	`if (res == SQLITE_OK) sqlite3_step(stmt);`
125	`sqlite3_finalize(stmt);`	131	`sqlite3_finalize(stmt);`
126	`return res; /* error on dup name? */`	132	`return res; /* error on dup name? */`
127	`}`	133	`}`
128		134
129	`/* copy a factor\|ordered table from a sdf(db) to another sdf(db) */`	135	`/* copy a factor\|ordered table from a sdf(db) to another sdf(db) */`
130	`int _copy_factor_levels2(const char factor_type, const char iname_src,`	136	`int _copy_factor_levels2(const char factor_type, const char iname_src,`
131	`const char colname_src, const char iname_dst, const char *colname_dst) {`	137	`const char colname_src, const char iname_dst, const char *colname_dst) {`
132	`sqlite3_stmt *stmt;`	138	`sqlite3_stmt *stmt;`
133	`int res;`	139	`int res;`
134	`res = _create_factor_table2(iname_dst, factor_type, colname_dst);`	140	`res = _create_factor_table2(iname_dst, factor_type, colname_dst);`
135	`if (res == SQLITE_OK) {`	141	`if (res == SQLITE_OK) {`
136	`sprintf(g_sql_buf[2], "insert into [%s].[%s %s] select * from [%s].[%s %s]",`	142	`sprintf(g_sql_buf[2], "insert into [%s].[%s %s] select * from [%s].[%s %s]",`
137	`iname_dst, factor_type, colname_dst, iname_src, factor_type,`	143	`iname_dst, factor_type, colname_dst, iname_src, factor_type,`
138	`colname_src);`	144	`colname_src);`
139	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);`	145	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, 0);`
140	`if (res == SQLITE_OK) res = sqlite3_step(stmt);`	146	`if (res == SQLITE_OK) res = sqlite3_step(stmt);`
141	`sqlite3_finalize(stmt);`	147	`sqlite3_finalize(stmt);`
142	`}`	148	`}`
143	`return res; /* error on dup name? */`	149	`return res; /* error on dup name? */`
144	`}`	150	`}`
145		151
146		152
147	`/* assumes stmt has col_cnt + 1 columns, col 0 is [row name]. returned SEXP is`	153	`/* assumes stmt has col_cnt + 1 columns, col 0 is [row name]. returned SEXP is`
148	`* not UNPROTECT-ed, user will have to do that. will create the names &`	154	`* not UNPROTECT-ed, user will have to do that. will create the names &`
149	`* attach factor infos */`	155	`* attach factor infos */`
150	`static SEXP _setup_df_sexp1(sqlite3_stmt stmt, const char iname,`	156	`static SEXP _setup_df_sexp1(sqlite3_stmt stmt, const char iname,`
151	`int col_cnt, int row_cnt, int *dup_indices) {`	157	`int col_cnt, int row_cnt, int *dup_indices) {`
152	`SEXP ret, names, value, class = R_NilValue;`	158	`SEXP ret, names, value, class = R_NilValue;`
153	`int i, type;`	159	`int i, type;`
154	`const char colname, coltype;`	160	`const char colname, coltype;`
155		161
156	`PROTECT(ret = NEW_LIST(col_cnt));`	162	`PROTECT(ret = NEW_LIST(col_cnt));`
157		163
158	`/* set up names. */`	164	`/* set up names. */`
159	`PROTECT(names = NEW_CHARACTER(col_cnt));`	165	`PROTECT(names = NEW_CHARACTER(col_cnt));`
160	`for (i = 0; i < col_cnt; i++) {`	166	`for (i = 0; i < col_cnt; i++) {`
161	`colname = sqlite3_column_name(stmt, i+1); /* +1 bec [row name is 0] */`	167	`colname = sqlite3_column_name(stmt, i+1); /* +1 bec [row name is 0] */`
162	`coltype = sqlite3_column_decltype(stmt, i+1);`	168	`coltype = sqlite3_column_decltype(stmt, i+1);`
163		169
164	`if (dup_indices[i] == 0) {`	170	`if (dup_indices[i] == 0) {`
165	`strcpy(g_sql_buf[1], colname);`	171	`strcpy(g_sql_buf[1], colname);`
166	`} else {`	172	`} else {`
167	`sprintf(g_sql_buf[1], "%s.%d", colname, dup_indices[i]);`	173	`sprintf(g_sql_buf[1], "%s.%d", colname, dup_indices[i]);`
168	`}`	174	`}`
169		175
170	`SET_STRING_ELT(names, i, mkChar(g_sql_buf[1]));`	176	`SET_STRING_ELT(names, i, mkChar(g_sql_buf[1]));`
171		177
172	`if (strcmp(coltype, "text") == 0) {`	178	`if (strcmp(coltype, "text") == 0) {`
173	`PROTECT(value = NEW_CHARACTER(row_cnt));`	179	`PROTECT(value = NEW_CHARACTER(row_cnt));`
174	`} else if (strcmp(coltype, "double") == 0) {`	180	`} else if (strcmp(coltype, "double") == 0) {`
175	`PROTECT(value = NEW_NUMERIC(row_cnt));`	181	`PROTECT(value = NEW_NUMERIC(row_cnt));`
176	`} else if (strcmp(coltype, "bit") == 0) {`	182	`} else if (strcmp(coltype, "bit") == 0) {`
177	`PROTECT(value = NEW_LOGICAL(row_cnt));`	183	`PROTECT(value = NEW_LOGICAL(row_cnt));`
178	`} else if (strcmp(coltype, "integer") == 0 \|\|`	184	`} else if (strcmp(coltype, "integer") == 0 \|\|`
179	`strcmp(coltype, "int") == 0) {`	185	`strcmp(coltype, "int") == 0) {`
180	`PROTECT(value = NEW_INTEGER(row_cnt));`	186	`PROTECT(value = NEW_INTEGER(row_cnt));`
181	`type = _get_factor_levels1(iname, colname, value);`	187	`type = _get_factor_levels1(iname, colname, value);`
182	`if (type == VAR_FACTOR) {`	188	`if (type == VAR_FACTOR) {`
183	`PROTECT(class = mkString("factor"));`	189	`PROTECT(class = mkString("factor"));`
184	`SET_CLASS(value, class);`	190	`SET_CLASS(value, class);`
185	`UNPROTECT(1);`	191	`UNPROTECT(1);`
186	`} else if (type == VAR_ORDERED) {`	192	`} else if (type == VAR_ORDERED) {`
187	`PROTECT(class = NEW_CHARACTER(2));`	193	`PROTECT(class = NEW_CHARACTER(2));`
188	`SET_STRING_ELT(class, 0, mkChar("ordered"));`	194	`SET_STRING_ELT(class, 0, mkChar("ordered"));`
189	`SET_STRING_ELT(class, 1, mkChar("factor"));`	195	`SET_STRING_ELT(class, 1, mkChar("factor"));`
190	`SET_CLASS(value, class);`	196	`SET_CLASS(value, class);`
191	`UNPROTECT(1);`	197	`UNPROTECT(1);`
192	`}`	198	`}`
193	`} else {`	199	`} else {`
194	`Rprintf("Error: not supported type %s for %s\n", coltype, colname);`	200	`Rprintf("Error: not supported type %s for %s\n", coltype, colname);`
195	`UNPROTECT(2); /* unprotect ret, names */`	201	`UNPROTECT(2); /* unprotect ret, names */`
196	`return R_NilValue;`	202	`return R_NilValue;`
197	`}`	203	`}`
198		204
199	`SET_VECTOR_ELT(ret, i, value);`	205	`SET_VECTOR_ELT(ret, i, value);`
200	`UNPROTECT(1); /* unprotect value */`	206	`UNPROTECT(1); /* unprotect value */`
201		207
202	`}`	208	`}`
203	`SET_NAMES(ret, names);`	209	`SET_NAMES(ret, names);`
204	`UNPROTECT(1); /* unprotect names only */`	210	`UNPROTECT(1); /* unprotect names only */`
205	`return ret;`	211	`return ret;`
206	`}`	212	`}`
207		213
208	`/* expected that 1st col of stmt is [row name], so we start with index 1 */`	214	`/* expected that 1st col of stmt is [row name], so we start with index 1 */`
209	`static int _add_row_to_df(SEXP df, sqlite3_stmt *stmt, int row, int ncols) {`	215	`static int _add_row_to_df(SEXP df, sqlite3_stmt *stmt, int row, int ncols) {`
210	`SEXP vec;`	216	`SEXP vec;`
211	`int type = -1, is_null, i;`	217	`int type = -1, is_null, i;`
212	`for (i = 1; i <= ncols; i++) {`	218	`for (i = 1; i <= ncols; i++) {`
213	`vec = VECTOR_ELT(df, i-1);`	219	`vec = VECTOR_ELT(df, i-1);`
214	`type = TYPEOF(vec);`	220	`type = TYPEOF(vec);`
215		221
216	`is_null = (sqlite3_column_type(stmt, row) == SQLITE_NULL);`	222	`is_null = (sqlite3_column_type(stmt, row) == SQLITE_NULL);`
217		223
218	`if (type == STRSXP \|\| type == CHARSXP) {`	224	`if (type == STRSXP \|\| type == CHARSXP) {`
219	`SET_STRING_ELT(vec, row, mkChar((char *)sqlite3_column_text(stmt, i)));`	225	`SET_STRING_ELT(vec, row, mkChar((char *)sqlite3_column_text(stmt, i)));`
220	`} else if (type == INTSXP) {`	226	`} else if (type == INTSXP) {`
221	`INTEGER(vec)[row] = sqlite3_column_int(stmt, i);`	227	`INTEGER(vec)[row] = sqlite3_column_int(stmt, i);`
222	`} else if (type == REALSXP) {`	228	`} else if (type == REALSXP) {`
223	`REAL(vec)[row] = sqlite3_column_double(stmt, i);`	229	`REAL(vec)[row] = sqlite3_column_double(stmt, i);`
224	`} else if (type == LGLSXP) {`	230	`} else if (type == LGLSXP) {`
225	`LOGICAL(vec)[row] = sqlite3_column_int(stmt, i);`	231	`LOGICAL(vec)[row] = sqlite3_column_int(stmt, i);`
226	`}`	232	`}`
227	`}`	233	`}`
228	`return type;`	234	`return type;`
229	`}`	235	`}`
230		236
231	`/* set ordinary df row name as numbers */`	237	`/* set ordinary df row name as numbers */`
232	`static void _set_rownames2(SEXP df) {`	238	`static void _set_rownames2(SEXP df) {`
233	`SEXP value = VECTOR_ELT(df, 0);`	239	`SEXP value = VECTOR_ELT(df, 0);`
234	`int len = LENGTH(value), i;`	240	`int len = LENGTH(value), i;`
235	`PROTECT(value = NEW_CHARACTER(len));`	241	`PROTECT(value = NEW_CHARACTER(len));`
236	`for (i = 0; i < len; i++) {`	242	`for (i = 0; i < len; i++) {`
237	`sprintf(g_sql_buf[2], "%d", i+1);`	243	`sprintf(g_sql_buf[2], "%d", i+1);`
238	`SET_STRING_ELT(value, i, mkChar(g_sql_buf[2]));`	244	`SET_STRING_ELT(value, i, mkChar(g_sql_buf[2]));`
239	`}`	245	`}`
240		246
241	`SET_ROWNAMES(df, value);`	247	`SET_ROWNAMES(df, value);`
242	`UNPROTECT(1);`	248	`UNPROTECT(1);`
243	`}`	249	`}`
244		250
245		251
246	`/****************************************************************************`	252	`/****************************************************************************`
247	`* SDF FUNCTIONS`	253	`* SDF FUNCTIONS`
248	`****************************************************************************/`	254	`****************************************************************************/`
249		255
250	`SEXP sdf_create_sdf(SEXP df, SEXP name) {`	256	`SEXP sdf_create_sdf(SEXP df, SEXP name) {`
251	`SEXP ret;`	257	`SEXP ret;`
252	`char *iname;`	258	`char *iname;`
253	`int namelen, res, i, j;`	259	`int namelen, res, i, j;`
254		260
255	`/* find free name, attach sdf, create sdf_attributes */`	261	`/* find free name, attach sdf, create sdf_attributes */`
256	`iname = _create_sdf_skeleton1(name, &namelen, FALSE);`	262	`iname = _create_sdf_skeleton1(name, &namelen, FALSE);`
257		263
258	`if (iname != NULL) {`	264	`if (iname != NULL) {`
259	`int sql_len, sql_len2;`	265	`int sql_len, sql_len2;`
260	`sqlite3_stmt *stmt;`	266	`sqlite3_stmt *stmt;`
261		267
262	`/* create sdf_data table */`	268	`/* create sdf_data table */`
263	`SEXP names = GET_NAMES(df), variable, levels, var_class;`	269	`SEXP names = GET_NAMES(df), variable, levels, var_class;`
264	`int ncols = GET_LENGTH(names), type, *types;`	270	`int ncols = GET_LENGTH(names), type, *types;`
265	`char col_name, class, *factor;`	271	`char col_name, class, *factor;`
266		272
267	`/* variables for adding rownames */`	273	`/* variables for adding rownames */`
268	`SEXP rownames;`	274	`SEXP rownames;`
269	`int nrows;`	275	`int nrows;`
270	`char *row_name;`	276	`char *row_name;`
271		277
272	`/* TODO: put constraints on table after inserting everything? */`	278	`/* TODO: put constraints on table after inserting everything? */`
273		279
274		280
275	`/*`	281	`/*`
276	`* create the create table and insert sql scripts by looping through`	282	`* create the create table and insert sql scripts by looping through`
277	`* the columns of df`	283	`* the columns of df`
278	`*/`	284	`*/`
279	`types = (int *)R_alloc(ncols, sizeof(int));`	285	`types = (int *)R_alloc(ncols, sizeof(int));`
280	`sql_len = sprintf(g_sql_buf[0], "create table [%s].sdf_data ([row name] text", iname);`	286	`sql_len = sprintf(g_sql_buf[0], "create table [%s].sdf_data ([row name] text", iname);`
281	`sql_len2 = sprintf(g_sql_buf[1], "insert into [%s].sdf_data values(?", iname);`	287	`sql_len2 = sprintf(g_sql_buf[1], "insert into [%s].sdf_data values(?", iname);`
282		288
283	`for (i = 0; i < ncols; i++) {`	289	`for (i = 0; i < ncols; i++) {`
284	`col_name = CHAR(STRING_ELT(names, i));`	290	`col_name = CHAR(STRING_ELT(names, i));`
285		291
286	`/* add column definition to the create table sql */`	292	`/* add column definition to the create table sql */`
287	`_expand_buf(0, sql_len+strlen(col_name)+10);`	293	`_expand_buf(0, sql_len+strlen(col_name)+10);`
288		294
289	`variable = _getListElement(df, col_name);`	295	`variable = _getListElement(df, col_name);`
290	`var_class = GET_CLASS(variable);`	296	`var_class = GET_CLASS(variable);`
291	`class = (var_class == R_NilValue) ? NULL : CHAR(STRING_ELT(var_class, 0));`	297	`class = (var_class == R_NilValue) ? NULL : CHAR(STRING_ELT(var_class, 0));`
292	`type = TYPEOF(variable);`	298	`type = TYPEOF(variable);`
293	`types[i] = type;`	299	`types[i] = type;`
294		300
295	`sql_len += sprintf(g_sql_buf[0]+sql_len, ", [%s] %s", col_name,`	301	`sql_len += sprintf(g_sql_buf[0]+sql_len, ", [%s] %s", col_name,`
296	`_get_column_type(class, type));`	302	`_get_column_type(class, type));`
297		303
298	`/* add handler to insert table sql */`	304	`/* add handler to insert table sql */`
299	`_expand_buf(1, sql_len+5);`	305	`_expand_buf(1, sql_len+5);`
300	`strcpy(g_sql_buf[1]+sql_len2, ",?");`	306	`strcpy(g_sql_buf[1]+sql_len2, ",?");`
301	`sql_len2 += 2;`	307	`sql_len2 += 2;`
302		308
303	`/* create separate table for factors decode */`	309	`/* create separate table for factors decode */`
304	`if (class != NULL && (strcmp(class, "factor") == 0 \|\| strcmp(class, "ordered") == 0)){`	310	`if (class != NULL && (strcmp(class, "factor") == 0 \|\| strcmp(class, "ordered") == 0)){`
305	`if (_create_factor_table2(iname, class, col_name))`	311	`if (_create_factor_table2(iname, class, col_name))`
306	`return R_NilValue; /* dup tbl name? */`	312	`return R_NilValue; /* dup tbl name? */`
307		313
308	`_sqlite_exec("begin");`	314	`_sqlite_exec("begin");`
309	`levels = GET_LEVELS(variable);`	315	`levels = GET_LEVELS(variable);`
310	`sprintf(g_sql_buf[2], "insert into [%s].[%s %s] values(?, ?);",`	316	`sprintf(g_sql_buf[2], "insert into [%s].[%s %s] values(?, ?);",`
311	`iname, class, col_name);`	317	`iname, class, col_name);`
312	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, NULL);`	318	`res = sqlite3_prepare(g_workspace, g_sql_buf[2], -1, &stmt, NULL);`
313	`if (_sqlite_error(res)) return R_NilValue; /* dup tbl name? */`	319	`if (_sqlite_error(res)) return R_NilValue; /* dup tbl name? */`
314		320
315	`for (j = 0; j < GET_LENGTH(levels); j++) {`	321	`for (j = 0; j < GET_LENGTH(levels); j++) {`
316	`sqlite3_reset(stmt);`	322	`sqlite3_reset(stmt);`
317	`factor = CHAR(STRING_ELT(levels, j));`	323	`factor = CHAR(STRING_ELT(levels, j));`
318	`sqlite3_bind_int(stmt, 1, j+1);`	324	`sqlite3_bind_int(stmt, 1, j+1);`
319	`sqlite3_bind_text(stmt, 2, factor, -1, SQLITE_STATIC);`	325	`sqlite3_bind_text(stmt, 2, factor, -1, SQLITE_STATIC);`
320	`sqlite3_step(stmt);`	326	`sqlite3_step(stmt);`
321	`}`	327	`}`
322	`sqlite3_finalize(stmt);`	328	`sqlite3_finalize(stmt);`
323	`_sqlite_exec("commit");`	329	`_sqlite_exec("commit");`
324	`}`	330	`}`
325	`}`	331	`}`
326		332
327	`_expand_buf(0,sql_len+35);`	333	`_expand_buf(0,sql_len+35);`
328	`sql_len += sprintf(g_sql_buf[0]+sql_len, ", primary key([row name]));");`	334	`sql_len += sprintf(g_sql_buf[0]+sql_len, ", primary key([row name]));");`
329	`/* sql_len += sprintf(g_sql_buf[0]+sql_len, ");"); */`	335	`/* sql_len += sprintf(g_sql_buf[0]+sql_len, ");"); */`
330	`res = _sqlite_exec(g_sql_buf[0]);`	336	`res = _sqlite_exec(g_sql_buf[0]);`
331	`if (_sqlite_error(res)) return R_NilValue; /* why? */`	337	`if (_sqlite_error(res)) return R_NilValue; /* why? */`
332		338
333	`/*`	339	`/*`
334	`* add the data in df to the sdf`	340	`* add the data in df to the sdf`
335	`*/`	341	`*/`
336	`rownames = getAttrib(df, R_RowNamesSymbol);`	342	`rownames = getAttrib(df, R_RowNamesSymbol);`
337	`nrows = GET_LENGTH(rownames);`	343	`nrows = GET_LENGTH(rownames);`
338		344
339	`_sqlite_error(_sqlite_exec("begin"));`	345	`_sqlite_error(_sqlite_exec("begin"));`
340	`sprintf(g_sql_buf[1]+sql_len2, ")");`	346	`sprintf(g_sql_buf[1]+sql_len2, ")");`
341	`res = sqlite3_prepare(g_workspace, g_sql_buf[1], -1, &stmt, NULL);`	347	`res = sqlite3_prepare(g_workspace, g_sql_buf[1], -1, &stmt, NULL);`
342		348
343	`for (i = 0; i < nrows; i++) {`	349	`for (i = 0; i < nrows; i++) {`
344	`sqlite3_reset(stmt);`	350	`sqlite3_reset(stmt);`
345		351
346	`/* since this is coming from a real dataframe, we're assured that`	352	`/* since this is coming from a real dataframe, we're assured that`
347	`* the rownames are unique */`	353	`* the rownames are unique */`
348	`if (IS_CHARACTER(rownames)) {`	354	`if (IS_CHARACTER(rownames)) {`
349	`row_name = CHAR(STRING_ELT(rownames, i));`	355	`row_name = CHAR(STRING_ELT(rownames, i));`
350	`if (row_name) / if not empty string */`	356	`if (row_name) / if not empty string */`
351	`sqlite3_bind_text(stmt, 1, row_name, strlen(row_name), SQLITE_STATIC);`	357	`sqlite3_bind_text(stmt, 1, row_name, strlen(row_name), SQLITE_STATIC);`
352	`else sqlite3_bind_int(stmt, 1, i);`	358	`else sqlite3_bind_int(stmt, 1, i);`
353	`} else if (IS_INTEGER(rownames)) {`	359	`} else if (IS_INTEGER(rownames)) {`
354	`sqlite3_bind_int(stmt, 1, INTEGER(rownames)[i]);`	360	`sqlite3_bind_int(stmt, 1, INTEGER(rownames)[i]);`
355	`} else sqlite3_bind_int(stmt, 1, i);`	361	`} else sqlite3_bind_int(stmt, 1, i);`
356		362
357	`for (j = 0; j < ncols; j++) {`	363	`for (j = 0; j < ncols; j++) {`
358	`variable = VECTOR_ELT(df, j);`	364	`variable = VECTOR_ELT(df, j);`
359	`switch(types[j]) {`	365	`switch(types[j]) {`
360	`case INTSXP :`	366	`case INTSXP :`
361	`sqlite3_bind_int(stmt, j+2, INTEGER(variable)[i]);`	367	`sqlite3_bind_int(stmt, j+2, INTEGER(variable)[i]);`
362	`break;`	368	`break;`
363	`case REALSXP:`	369	`case REALSXP:`
364	`sqlite3_bind_double(stmt, j+2, REAL(variable)[i]);`	370	`sqlite3_bind_double(stmt, j+2, REAL(variable)[i]);`
365	`break;`	371	`break;`
366	`case CHARSXP:`	372	`case CHARSXP:`
367	`case STRSXP:`	373	`case STRSXP:`
368	`sqlite3_bind_text(stmt, j+2, CHAR_ELT(variable, i), -1, SQLITE_STATIC);`	374	`sqlite3_bind_text(stmt, j+2, CHAR_ELT(variable, i), -1, SQLITE_STATIC);`
369	`}`	375	`}`
370	`/* TODO: handle NA's & NULL's */`	376	`/* TODO: handle NA's & NULL's */`
371	`}`	377	`}`
372		378
373	`res = sqlite3_step(stmt);`	379	`res = sqlite3_step(stmt);`
374	`if (res != SQLITE_DONE) {`	380	`if (res != SQLITE_DONE) {`
375	`_sqlite_exec("ROLLBACK");`	381	`_sqlite_exec("ROLLBACK");`
376	`sqlite3_finalize(stmt);`	382	`sqlite3_finalize(stmt);`
377	`Rprintf("SQLITE ERROR: %s\n", sqlite3_errmsg(g_workspace));`	383	`Rprintf("SQLITE ERROR: %s\n", sqlite3_errmsg(g_workspace));`
378	`return R_NilValue; /* why? */`	384	`return R_NilValue; /* why? */`
379	`}`	385	`}`
380	`}`	386	`}`
381		387
382	`sqlite3_finalize(stmt);`	388	`sqlite3_finalize(stmt);`
383	`_sqlite_error(_sqlite_exec("COMMIT"));`	389	`_sqlite_error(_sqlite_exec("COMMIT"));`
384		390
385	`/* create a new object representing the sdf */`	391	`/* create a new object representing the sdf */`
386	`ret = _create_sdf_sexp(iname);`	392	`ret = _create_sdf_sexp(iname);`
387		393
388	`} else {`	394	`} else {`
389	`Rprintf("ERROR: unable to create a sqlite data frame.\n");`	395	`Rprintf("ERROR: unable to create a sqlite data frame.\n");`
390	`ret = R_NilValue;`	396	`ret = R_NilValue;`
391	`}`	397	`}`
392		398
393	`return ret;`	399	`return ret;`
394	`}`	400	`}`
395		401
396	`SEXP sdf_get_names(SEXP sdf) {`	402	`SEXP sdf_get_names(SEXP sdf) {`
397	`char *iname;`	403	`char *iname;`
398	`iname = SDF_INAME(sdf);`	404	`iname = SDF_INAME(sdf);`
399	`if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;`	405	`if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;`
400		406
401	`int len;`	407	`int len;`
402	`len = sprintf(g_sql_buf[0], "select * from [%s].sdf_data;", iname);`	408	`len = sprintf(g_sql_buf[0], "select * from [%s].sdf_data;", iname);`
403		409
404	`sqlite3_stmt *stmt;`	410	`sqlite3_stmt *stmt;`
405	`int res, i;`	411	`int res, i;`
406		412
407	`res = sqlite3_prepare(g_workspace, g_sql_buf[0], len, &stmt, NULL);`	413	`res = sqlite3_prepare(g_workspace, g_sql_buf[0], len, &stmt, NULL);`
408	`if (_sqlite_error(res)) return R_NilValue;`	414	`if (_sqlite_error(res)) return R_NilValue;`
409		415
410	`SEXP ret;`	416	`SEXP ret;`
411		417
412	`len = sqlite3_column_count(stmt)-1;`	418	`len = sqlite3_column_count(stmt)-1;`
413	`PROTECT(ret = NEW_CHARACTER(len));`	419	`PROTECT(ret = NEW_CHARACTER(len));`
414		420
415	`for (i = 0; i < len; i++) {`	421	`for (i = 0; i < len; i++) {`
416	`SET_STRING_ELT(ret, i, mkChar(sqlite3_column_name(stmt, i+1)));`	422	`SET_STRING_ELT(ret, i, mkChar(sqlite3_column_name(stmt, i+1)));`
417	`}`	423	`}`
418		424
419	`sqlite3_finalize(stmt);`	425	`sqlite3_finalize(stmt);`
420	`UNPROTECT(1);`	426	`UNPROTECT(1);`
421	`return ret;`	427	`return ret;`
422	`}`	428	`}`
423		429
424	`SEXP sdf_get_length(SEXP sdf) {`	430	`SEXP sdf_get_length(SEXP sdf) {`
425	`char *iname;`	431	`char *iname;`
426	`int len;`	432	`int len;`
427	`sqlite3_stmt *stmt;`	433	`sqlite3_stmt *stmt;`
428	`int res;`	434	`int res;`
429		435
430	`iname = CHAR(STRING_ELT(_getListElement(sdf, "iname"),0));`	436	`iname = CHAR(STRING_ELT(_getListElement(sdf, "iname"),0));`
431	`sprintf(g_sql_buf[0], "select * from [%s].sdf_data;", iname);`	437	`sprintf(g_sql_buf[0], "select * from [%s].sdf_data;", iname);`
432	`if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;`	438	`if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;`
433		439
434	`res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, NULL);`	440	`res = sqlite3_prepare(g_workspace, g_sql_buf[0], -1, &stmt, NULL);`
435	`if (_sqlite_error(res)) return R_NilValue;`	441	`if (_sqlite_error(res)) return R_NilValue;`
436		442
437	`len = sqlite3_column_count(stmt)-1;`	443	`len = sqlite3_column_count(stmt)-1;`
438	`sqlite3_finalize(stmt);`	444	`sqlite3_finalize(stmt);`
439	`return ScalarInteger(len);`	445	`return ScalarInteger(len);`
440	`}`	446	`}`
441		447
442	`/* get row count */`	448	`/* get row count */`
443	`SEXP sdf_get_row_count(SEXP sdf) {`	449	`SEXP sdf_get_row_count(SEXP sdf) {`
444	`char *iname = CHAR(STRING_ELT(_getListElement(sdf, "iname"),0));`	450	`char *iname = CHAR(STRING_ELT(_getListElement(sdf, "iname"),0));`
445	`if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;`	451	`if (!USE_SDF1(iname, TRUE, FALSE)) return R_NilValue;`
446		452
447	`int nrow;`	453	`int nrow;`
448	`SEXP ret;`	454	`SEXP ret;`
449		455
450	`nrow = _get_row_count2(iname, TRUE);`	456	`nrow = _get_row_count2(iname, TRUE);`
451		457
452	`if (nrow < 1) ret = R_NilValue;`	458	`if (nrow < 1) ret = R_NilValue;`
453	`else ret = ScalarInteger(nrow);`	459	`else ret = ScalarInteger(nrow);`
454		460
455	`return ret;`	461	`return ret;`
456	`}`	462	`}`
457		463
458		464
459	`SEXP sdf_import_table(SEXP _filename, SEXP _name, SEXP _sep, SEXP _quote,`	465	`SEXP sdf_import_table(SEXP _filename, SEXP _name, SEXP _sep, SEXP _quote,`
460	`SEXP _rownames, SEXP _colnames) {`	466	`SEXP _rownames, SEXP _colnames) {`
461	`char *filename = CHAR_ELT(_filename, 0);`	467	`char *filename = CHAR_ELT(_filename, 0);`
462	`FILE *f = fopen(filename, "r");`	468	`FILE *f = fopen(filename, "r");`
463		469
464	`if (f == NULL) {`	470	`if (f == NULL) {`
465	`Rprintf("Error: File %s does not exist.", filename);`	471	`Rprintf("Error: File %s does not exist.", filename);`
466	`return R_NilValue;`	472	`return R_NilValue;`
467	`}`	473	`}`
468		474
469	`/char sep = CHAR_ELT(_sep, 0), *quote = CHAR_ELT(_quote,0);`	475	`/char sep = CHAR_ELT(_sep, 0), *quote = CHAR_ELT(_quote,0);`
470	`char name = CHAR_ELT(_name, 0);/`	476	`char name = CHAR_ELT(_name, 0);/`
471		477
472	`/* create the table */`	478	`/* create the table */`
473	`/* insert the stuffs */`	479	`/* insert the stuffs */`
474	`/* register to workspace */`	480	`/* register to workspace */`
475		481
476	`return R_NilValue;`	482	`return R_NilValue;`
477	`}`	483	`}`
478		484
479	`SEXP sdf_get_index(SEXP sdf, SEXP row, SEXP col, SEXP new_sdf) {`	485	`SEXP sdf_get_index(SEXP sdf, SEXP row, SEXP col, SEXP new_sdf) {`
480	`SEXP ret = R_NilValue;`	486	`SEXP ret = R_NilValue;`
481	`char *iname = SDF_INAME(sdf);`	487	`char *iname = SDF_INAME(sdf);`
482	`sqlite3_stmt *stmt;`	488	`sqlite3_stmt *stmt;`
483	`int buflen = 0, idxlen, col_cnt, row_cnt, index;`	489	`int buflen = 0, idxlen, col_cnt, row_cnt, index;`
484	`int i, j,res;`	490	`int i, j,res;`
485	`int col_indices, col_index_len, dup_indices;`	491	`int col_indices, col_index_len, dup_indices;`
486	`int row_index_len = 0, force_new_df = LOGICAL(new_sdf)[0];`	492	`int row_index_len = 0, force_new_df = LOGICAL(new_sdf)[0];`
487	`const char *colname;`	493	`const char *colname;`
488		494
489	`/* dup_indices is used to handle when same column chosen more than once.`	495	`/* dup_indices is used to handle when same column chosen more than once.`
490	`* e.g. iris[,c(1,1)] have names Sepal.Length, Sepal.Length.1 *`	496	`* e.g. iris[,c(1,1)] have names Sepal.Length, Sepal.Length.1 *`
491	`* col_indices is used to store the column index of selected columns`	497	`* col_indices is used to store the column index of selected columns`
492	`* in the sdf_data table */`	498	`* in the sdf_data table */`

The R Project SVN R-packages

(root)/trunk/SQLiteDF/src/sqlite_dataframe.c – Rev 3473 → 3684